https://bugzilla.redhat.com/show_bug.cgi?id=1112864 --- Comment #10 from Dave Love <d.love@xxxxxxxxxxxxxxx> --- (In reply to Dominik 'Rathann' Mierzejewski from comment #3) > > Would you accept a patch for EPEL6? > > Of course. I'll most likely keep it in that branch only, though. I'll attach the patch I've built with, though it's messy. > > It would probably be useful to package versions with other optimizations, > > like sandybridge in our case, but I wasn't sure how best to do it and > > couldn't find any examples. I wonder if there should be a policy on that. > > atlas and qtwebkit do something like that, for example, but only for i686 > with and without SSE2 support. I hadn't realized atlas was doing that now -- I use openblas when possible. Actually atlas has x86_64 sse2/sse3 versions. I'll look at doing something similar with ELPA unless there's a good reason not to. > Unfortunately, hwcap support is completely > undocumented (see my 6 year old bug: > https://sourceware.org/bugzilla/show_bug.cgi?id=7100) and no hwcaps are > defined on x86_64 (check /usr/include/bits/hwcap.h on your system). Yes, I'd ruled it out long ago, even though it seems to be just what's needed on a typical heterogeneous HPC system. Linux fails to do several things right that would really help HPC... -- You are receiving this mail because: You are on the CC list for the bug. You are always notified about changes to this product and component _______________________________________________ package-review mailing list package-review@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/package-review